Page 130 - Kaleidoscope Academic Conference Proceedings 2020

2020 ITU Kaleidoscope Academic Conference

performance of the proposed techniques show that the [4] H. Zhang, C.D. Wang, and J.H. Lai, “Topic detection
proposed technique works better than the existing techniques in instant messages,” 13th International Conference on
in terms of efficiency as well as effectiveness in terms of Machine Learning and Applications, Detroit, MI, pp.
accuracy. 219-224, 2014. doi:10.1109/ICMLA.2014.41.
[5] F. Iqbal, B.C.M. Fung, M. Debbabi, R. Batool, and A.
It is suggested that the work presented in this paper can be Marrington, “Wordnet-based criminal networks
further developed in the following directions. mining for cybercrime investigation,” IEEE Access,

vol. 7, pp. 22740-22755, 2019.
• A better spell checker to be incorporated: A better and doi:10.1109/ACCESS.2019.2891694.
dedicated spelling mistake-handling component using an
LSTM that is capable of capturing the phonetics, as well as [6] I. Song, and J. Diederich, “Intention extraction from
the context, when understanding the words for translating text messages,” International Conference on Neural
them into English text. Information Processing, Sydney, Australia, pp.330-
337, 2010. doi:10.1007/978-3-642-17537-4-41.
• Multi-language support: A dedicated module can be [7] G. Inches, and F. Crestani, “Online conversation
implemented that is capable of translating different languages mining for author characterization and topic
into English and the normal procedure can continue to identification,” 4th Workshop on Information and
classify the messages as appropriate and inappropriate. knowledge Management for Ph.D. Students, Glasgow,
UK, pp.19–26, 2011. doi:10.1145/2065003.2065009
• Handling multiple coded texts: Sometimes users mix
words from multiple languages within the same message. A [8] Y. Chen, Y. Yang, H. Zhang, H. Zhu and F. Tian, “A
special module can be created that can detect the words from topic detection method based on semantic dependency
other languages but written using English script and distance and PLSA,” IEEE 16th International
translating them. Conference on Computer Supported Cooperative Work
in Design (CSCWD), Wuhan, China, pp. 703-708,
• Supporting graphics: A special module to handle images 2012. doi:10.1109/CSCWD.2012.6221895.
and videos to be developed. [9] Nltk.org, “WordNet interface,” [online] Available at:
http://www.nltk.org/howto/wordnet.html, Accessed on
• Detecting other fraudulent intentions: Improving the Jan 15, 2020.
system to detect other types of fraudulent intentions including
hate speech, promoting terrorism and financial and non- [10] F. Pedregosa et al., “Scikit-learn: Machine learning in
financial frauds. Python,” Journal of Machine Learning Research,
vol.12, pp.2825-2830, 2011.
REFERENCES [11] L. Buitinck et al., “API design for machine learning
software: Experiences from the scikit-learn project,”
[1] T. Dodds, “Reporting with WhatsApp: Mobile chat European Conference on Machine Learning and
applications’ impact on journalistic practices,” Digital Principles and Practices of Knowledge Discovery in
Journalism, vol. 7, no. 6, pp. 725-745, 2019. doi: Databases, Prague, Czech Republic, pp.2825-2830,
10.1080/21670811.2019.1592693 2013.
[2] L. Cilliers and W. Chinyamurindi, “Perceptions of [12] P.H. Adams, and C.H. Martell, “Topic detection and
cyber bullying in primary and secondary schools extraction in chat,” IEEE International Conference on
among student teachers in the Eastern Cape Province Semantic Computing, Santa Clara, CA, USA, pp. 581-
of South Africa,” Electronic Journal of Information 588, 2008. doi:10.1109/ICSC.2008.61.
Systems in Developing Countries, vol. 86, no. 4, pp. 1-
10, 2020. doi: 10.1002/isd2.12131 [13] Anon, “Word vectors and semantic similarity,”
[online] Available at: https://spacy.io/usage/vectors-
[3] H. Dong, S.C. Hui, and Y. He, “Structural analysis of similarity, Accessed on Jan 15, 2020.
chat messages for topic detection,” Online Information
Review, vol.30 no.5, pp. 496-516, 2006. doi:
10.1108/14684520610706398.

– 72 –